Clinical Question

In adults with focal or multifocal drug-resistant epilepsy who have already failed vagus nerve stimulation, does bilateral deep brain stimulation of the anterior nucleus of the thalamus (ANT-DBS) reduce severe seizure frequency at 12 months compared with continued best medical treatment (including VNS)?

Inclusion Criteria

• Adults with focal or multifocal seizures (with or without focal-to-bilateral tonic–clonic seizures) ineligible for resective surgery or in whom resective surgery had failed
• Intractable seizures for ≥4 years
• ≥4 severe seizures/month for ≥3 months during the baseline period (minimum 12 seizures over 3 months)
• Severe seizures defined as Chalfont types B, C, or D on the modified Chalfont Scale (type A auras alone excluded)
• Failure of ≥2 antiseizure medications used for ≥2 years
• Failure of vagus nerve stimulation after ≥2 years of stimulation
• Written informed consent

Exclusion Criteria

• Generalized epilepsy
• Focal aware seizures with only auras (Chalfont type A only)
• Contraindication to DBS (vascular malformation, hemostatic disorder)
• Contraindication to MRI
• Evolutive disease (e.g., tumor)
• Suicide risk within the previous 6 months (Montgomery–Åsberg Depression Rating Scale item #10 ≥2)
• Women at risk of pregnancy during the study

Baseline Characteristics

Arms

Outcomes

Criticisms

• Small sample size (n=61) and open-label design; no sham stimulation control (patients aware of allocation), which may have introduced ascertainment bias in patient/family seizure diaries — the primary outcome data source.
• Underpowered — sample-size calculation assumed only 10% BMT responder rate (from Rheims 2008); observed BMT responder rate was ~27%, plausibly inflated by enhanced follow-up/diary attention, leaving the study without power to demonstrate the anticipated between-group difference.
• Primary endpoint (≥50% responder rate) was not met (p=.14), and the between-group median seizure-reduction comparison was also not significant (p=.09) — reported as a 'potential benefit' rather than superiority.
• Sex imbalance (72% male DBS vs 37% BMT) was not part of the minimization algorithm; post hoc adjustment showed no sex effect but this weakens confidence.
• IQ, processing speed, and perceptual reasoning were significantly higher in the BMT arm at baseline; temporal-onset seizures were more common in the BMT arm (p=.07) — possible cofounders.
• Reliance on patient/family-completed seizure diaries and the modified Chalfont severity scale (rather than continuous EEG/video) is subjective and prone to reporting bias, especially in an unblinded trial.
• Only 79% of assessed patients (of the 48 with imaging data, so 79% of a subset) had ≥2 contacts in the ANT — heterogeneous lead placement may have diluted effect.
• Post hoc analyses (median % change, subgroup by seizure onset) were exploratory; no adjustment for multiple comparisons.
• No significant improvement in quality of life on QOLIE-31 or any subscore — clinically the most patient-relevant outcome — despite within-arm seizure reductions.
• BDI depression 'improvement' finding may be misleading: baseline scores were already low (median 6/63) in both arms and 12-month median scores were similar (4 vs 4); the 88% vs 55% percentage-improved metric may reflect regression-to-the-mean rather than a true antidepressant effect.
• Higher SAE burden with DBS (45 vs 19 at 1 year) including 3 lead migrations, 3 infections needing explant, and 1 HSV encephalitis reactivation — non-trivial harms in an already high-risk population.
• Recruitment took 5 years across 14 centers for only 61 patients — likely reflects strict eligibility (VNS-failure required by French reimbursement) and limits generalizability outside France.

Funding

French Ministry of Health, Direction Générale de l'Offre de Soins (DGOS); Grenoble Alpes University Hospital served as sponsor/promoter. Open-access publication funded by COUPERIN CY26.